# HG changeset patch
# User ali <ali@juiblex.co.uk>
# Date 1327708791 0
# Node ID cf332d440466601401a5a8ce21918dee775cec27
# Parent  721e468c10f34914f9b79eb346e13dc95ddfdd27
Add a testcase for embedded LFs

diff -r 721e468c10f3 -r cf332d440466 test/compatibility/Makefile.am
--- a/test/compatibility/Makefile.am	Fri Jan 27 21:40:35 2012 +0000
+++ b/test/compatibility/Makefile.am	Fri Jan 27 23:59:51 2012 +0000
@@ -2,6 +2,6 @@
 TESTS=missing-space.tst spaced-punctuation.tst html-tag.tst html-symbol.tst \
 	spaced-doublequote.tst mismatched-quotes.tst he-be.tst digits.tst \
 	extra-period.tst ellipsis.tst short-line.tst abbreviation.tst \
-	example.tst non-ascii.tst
+	example.tst non-ascii.tst embedded-lf.tst
 
 dist_pkgdata_DATA=$(TESTS)
diff -r 721e468c10f3 -r cf332d440466 test/compatibility/embedded-lf.tst
--- /dev/null	Thu Jan 01 00:00:00 1970 +0000
+++ b/test/compatibility/embedded-lf.tst	Fri Jan 27 23:59:51 2012 +0000
@@ -0,0 +1,10 @@
+**************** INPUT ****************
+Katherine was assailed by a sudden doubt. Had she mailed that letter? Yes,␊she was certain of that. She had run out to the mail box at ten o'clock
+at night especially to mail it. What had gone wrong? Why wasn't there␊someone to meet her?
+**************** EXPECTED ****************
+
+Katherine was assailed by a sudden doubt. Had she mailed that letter? Yes,
+    Line 1 - No CR?
+
+at night especially to mail it. What had gone wrong? Why wasn't there
+    Line 3 - No CR?
diff -r 721e468c10f3 -r cf332d440466 test/harness/testcase.c
--- a/test/harness/testcase.c	Fri Jan 27 21:40:35 2012 +0000
+++ b/test/harness/testcase.c	Fri Jan 27 23:59:51 2012 +0000
@@ -3,10 +3,6 @@
 #include <string.h>
 #include <unistd.h>
 #include <errno.h>
-#ifdef WIN32
-#include <io.h>
-#endif
-#include <fcntl.h>
 #include <bl/bl.h>
 #include "testcase.h"
 
@@ -15,86 +11,12 @@
     return g_quark_from_static_string("testcase-error-quark");
 }
 
-#if !HAVE_MKSTEMP
 /*
- * An insecure implementation of mkstemp(), for those platforms that
- * don't support it.
+ * As write(), but with error handling.
  */
-int mkstemp(char *template)
+static size_t write_file(int fd,const char *buf,size_t count,GError **error)
 {
-    int fd;
-    char *s;
-    for(;;)
-    {
-	s=g_strdup(template);
-	mktemp(s);
-	if (!*s)
-	{
-	    errno=EEXIST;
-	    g_free(s);
-	    return -1;
-	}
-	fd=open(s,O_RDWR|O_CREAT|O_EXCL,0600);
-	if (fd>0)
-	{
-	    strcpy(template,s);
-	    g_free(s);
-	    return fd;
-	}
-	else
-	    g_free(s);
-    }
-}
-#endif	/* !HAVE_MKSTEMP */
-
-/*
- * As write(), but always convert NL to CR NL.
- */
-static size_t write_text(int fd,const char *buf,size_t count,GError **error)
-{
-    size_t i;
-    FILE *fp;
-    fd=dup(fd);
-    if (fd<0)
-	return -1;
-#ifdef WIN32
-    if (_setmode(fd,_O_BINARY)<0)
-    {
-	close(fd);
-	g_set_error(error,G_FILE_ERROR,g_file_error_from_errno(errno),
-	  "Failed to set mode of bookloupe input file to binary: %s",
-	  g_strerror(errno));
-	return -1;
-    }
-#endif
-    fp=fdopen(fd,"wb");
-    if (!fp)
-    {
-	close(fd);
-	g_set_error(error,G_FILE_ERROR,g_file_error_from_errno(errno),
-	  "Failed to open stream to bookloupe input file: %s",
-	  g_strerror(errno));
-	return -1;
-    }
-    for(i=0;i<count;i++)
-    {
-	if (buf[i]=='\n')
-	    if (putc('\r',fp)==EOF)
-	    {
-		g_set_error(error,G_FILE_ERROR,g_file_error_from_errno(errno),
-		  "Error writing bookloupe input file: %s",g_strerror(errno));
-		(void)fclose(fp);
-		return -1;
-	    }
-	if (putc(buf[i],fp)==EOF)
-	{
-	    g_set_error(error,G_FILE_ERROR,g_file_error_from_errno(errno),
-	      "Error writing bookloupe input file: %s",g_strerror(errno));
-	    (void)fclose(fp);
-	    return -1;
-	}
-    }
-    if (fclose(fp))
+    if (write(fd,buf,count)<count)
     {
 	g_set_error(error,G_FILE_ERROR,g_file_error_from_errno(errno),
 	  "Error writing bookloupe input file: %s",g_strerror(errno));
@@ -104,20 +26,30 @@
 }
 
 /*
- * Return the length (in bytes) or any common prefix between s1 and s2.
+ * Return the length (in bytes) of any common prefix between s1 and s2.
+ * The returned length will always represent an exact number of characters.
  */
 size_t common_prefix_length(const char *s1,const char *s2)
 {
-    size_t i;
-    for(i=0;s1[i] && s2[i] && s1[i]==s2[i];i++)
-	;
-    return i;
+    gunichar c1,c2;
+    const char *s=s1;
+    while(*s1 && *s2)
+    {
+	c1=g_utf8_get_char(s1);
+	c2=g_utf8_get_char(s2);
+	if (c1!=c2)
+	    break;
+	s1=g_utf8_next_char(s1);
+	s2=g_utf8_next_char(s2);
+    }
+    return s1-s;
 }
 
 void print_unexpected(const char *unexpected,gsize differs_at)
 {
     int col;
-    const char *endp,*bol;
+    gunichar c;
+    const char *endp,*bol,*s;
     GString *string;
     endp=strchr(unexpected+differs_at,'\n');
     if (!endp)
@@ -128,27 +60,64 @@
 	bol++;
     else
 	bol=string->str;
-    col=differs_at-(bol-string->str);
+    col=0;
+    s=bol;
+    endp=string->str+differs_at;
+    while(s<endp)
+    {
+	c=g_utf8_get_char(s);
+	s=g_utf8_next_char(s);
+	if (c=='\t')
+	    col=(col&~7)+8;
+	else if (g_unichar_iswide(c))
+	    col+=2;
+	else if (!g_unichar_iszerowidth(c))
+	    col++;
+    }
     fprintf(stderr,"%s\n%*s^\n",string->str,col,"");
     g_string_free(string,TRUE);
 }
 
+/*
+ * Replace \n with \r\n and U+240A (visible symbol for LF) with \n
+ */
+char *unix2dos(const char *text)
+{
+    gunichar c;
+    const gunichar visible_lf=0x240A;
+    GString *string;
+    string=g_string_new(NULL);
+    while(*text)
+    {
+	c=g_utf8_get_char(text);
+	text=g_utf8_next_char(text);
+	if (c=='\n')
+	    g_string_append(string,"\r\n");
+	else if (c==visible_lf)
+	    g_string_append_c(string,'\n');
+	else
+	    g_string_append_unichar(string,c);
+    }
+    return g_string_free(string,FALSE);
+}
+
 gboolean spawn_bootloupe(const char *encoding,const char *standard_input,
   char **standard_output,char **filename,GError **error)
 {
     gboolean r;
     int fd,exit_status;
     size_t n,pos,offset;
-    FILE *fp;
     char input[]="TEST-XXXXXX";
     char *command[3];
-    char *output,*s;
+    char *output,*s,*t;
     GError *tmp_err=NULL;
     if (standard_input)
     {
 	if (encoding)
 	{
-	    s=g_convert(standard_input,-1,encoding,"UTF-8",NULL,&n,&tmp_err);
+	    t=unix2dos(standard_input);
+	    s=g_convert(t,-1,encoding,"UTF-8",NULL,&n,&tmp_err);
+	    g_free(t);
 	    if (!s)
 	    {
 		g_propagate_prefixed_error(error,tmp_err,
@@ -158,7 +127,7 @@
 	}
 	else
 	{
-	    s=g_strdup(standard_input);
+	    s=unix2dos(standard_input);
 	    n=strlen(s);
 	}
     }
@@ -167,8 +136,8 @@
 	n=0;
 	s=NULL;
     }
-    fd=mkstemp(input);
-    if (n && write_text(fd,s,n,error)!=n)
+    fd=g_mkstemp(input);
+    if (n && write_file(fd,s,n,error)!=n)
     {
 	g_free(s);
 	close(fd);
diff -r 721e468c10f3 -r cf332d440466 test/harness/testcaseio.c
--- a/test/harness/testcaseio.c	Fri Jan 27 21:40:35 2012 +0000
+++ b/test/harness/testcaseio.c	Fri Jan 27 23:59:51 2012 +0000
@@ -39,7 +39,7 @@
 	else if (!testcase->expected && !strcmp(tag,"EXPECTED"))
 	    testcase->expected=g_strdup(text);
 	else if (!testcase->encoding && !strcmp(tag,"ENCODING"))
-	    testcase->encoding=g_strdup(text);
+	    testcase->encoding=g_strchomp(g_strdup(text));
 	else
 	{
 	    fprintf(stderr,"%s: Not a valid testcase (%s)\n",filename,tag);